Not Applicable
Not Applicable
The technical field on this invention relates to fluid handling in a plurality of reservoirs connected in series. This invention is a division of U.S. Pat. Application Ser. No. 09/655,296 filed on Sep. 5, 2000, now U.S. Pat. No. 6,422,263, entitled xe2x80x9cNested and Cascaded Variable Bias Feedforward and Feedback Flow And Level Control Systemxe2x80x9d which is hereby incorporated by reference. This invention involves innovative changes to the uses and configurations of the traditional proportional only (P), and proportional/integral/derivative control algorithms (PID) as they have been applied to maintain system demand and minimize disturbances to the critical levels within a process. Although this invention has been birthed out of needs specifically attributable to the water treatment industry as recognized by the applicant pursuant to over 20 years of process control experience in the field, the invention can be applied to all other liquid handling processes where demand fluctuates and reservoir or vessel level is desired to be maintained at a specific level.
Please refer to FIG. 1 Prior Art for a general simplified layout of a typical water treatment facility. This drawing shows pumped raw water influent to sedimentation basins, then flowing by gravity to a common flume which hydraulically links multiple sand filters. Each filter effluent then flows by gravity to an effluent clearwell from which water is pumped to maintain system demand.
The water treatment industry has long recognized the importance of minimizing filter flow rate fluctuations to minimize the breakthrough of turbidity (i.e. trapped sediment within the filter media bed) to the effluent clearwell to insure a clean water supply. Thus, different control schemes have been utilized by the industry to minimize water level (i.e. flume level) fluctuations in gravity flow sand filters. This is due to the fact that head pressure across the media bed varies with fluctuating level and since flow is proportional to the square root of the differential pressure across the filter, flow rate also fluctuates. Referring to FIG. 2 Prior Art depicting the relationship of flow rate to differential pressure, for a given effluent valve position the effect of pressure differential (as related to filter water level) is even more significant at lower flow rates.
To further aggravate the situation, as trapped sediment builds up within the media bed during the filter run period (filter run period defined as the period of time filtering occurs between filter backwashes) the hydraulic head pressure decreases, thus requiring the filter effluent valve to open further to maintain a set flow rate. Since filters are backwashed one at a time on a staggered basis to minimize disruption to the water plant""s water production capability, the head loss for a given filter will not be the same as any other and can vary by as much as four to five feet or more.
The scenario in which the quality of filtered water could be maximized would be where plant effluent demand would remain constant, filter level would remain constant, and the positions of the filter effluent valves would gradually increase over the course of the filter run to maintain a constant filter flow with increasing head loss across the filter media bed during the filter run. This scenario however is totally unrealistic since plant effluent demand does change and filters must be backwashed. Thus, the challenge is one where filtered water quality must be optimized with plant demand and other system disturbances, which requires filter flow rates to change as little as and as smoothly as possible. Based on fluctuating flume level with varying degrees of head loss as these variables relate to flow as described previously, if filter flow rates can be made to follow system demand while maintaining constant flume level, not only is filtered water quality optimized with plant demand, but the water production calculation data required by the regulatory agencies such as filter load rates are much more consistent, realistic and accurate.
The problem, however, becomes more complicated because at this point, only the first phase of water production and associated levels and flows has been discussed. Ultimately, filtered water must pass to the effluent clearwell, which is used as a storage medium for plant effluent pumping. As such, the water treatment industry has long recognized the need to minimize clearwell water level fluctuations in order to be responsive to system demand and various control schemes have been utilized to accomplish this. Further, the trend of state governments responding to the increased restrictions mandated by the Federal EPA for water quality has been to consider the effluent clearwell as the chemical contact chamber for chemical post treatment. This consideration requires the effluent clearwell to be maintained within certain set levels, below which fines may be imposed. This is so because it is recognized that contact time of the water to be treated with the chemicals is essential to the bonding process and the ultimate effectiveness of the chemical treatment, and that contact time increases with higher and reasonably steady state set levels. Similarly as for flume level, the water production calculation data required by the regulatory agencies such as chemical contact time is much more consistent, realistic, accurate, and accepted with a reasonably steady state clearwell level.
Further to the issue of responding to the increased requirements of the regulatory agencies, the credibility of the computer-based historical data collection and reporting systems concerning the provision of meaningfully accurate and consistent report data often times is a problem for municipalities. Since all computer-based data collection systems monitor process variable signal data by sampling techniques, accuracy decreases as these variables fluctuate, and this is again another reason to optimize the responsiveness of the water plant to system demand while minimizing the fluctuations in process level and flow variables. Inherent inaccuracies with these data collection systems are further aggravated when there are calibration errors and discrepancies amongst the various flow measuring devices that go undetected for indeterminate amounts of time.
Various control schemes have been implemented over the years in an attempt to control both flume level and clearwell level for the reasons mentioned previously. These schemes have attempted to address the challenge of maintaining these levels in a situation where the hydraulic design of the water plant requires the control aspects of these levels to be in competition with one another. These control schemes are herein described with their flaws.
Refer to the control scheme for flume level control depicted on FIG. 3 Prior Art. This configuration for flume level control utilizes dedicated proportional/integral/derivative (PID) flow controllers for each filter. A single proportional only (P) direct acting level controller output is cascaded to the set point inputs of the various PID reverse acting filter effluent flow controllers.
Referring to FIG. 4 Prior Art, shows the output responses of the direct acting proportional only flume level controller shown in FIG. 3 based on example gain values of 1 (waveform xe2x80x9cDxe2x80x9d) and 2 (waveform xe2x80x9cCxe2x80x9d), a bias constant of approximately 64% (waveform xe2x80x9cFxe2x80x9d), and a flume level set point of approximately 84% (waveform xe2x80x9cBxe2x80x9d). The typical proportional only filter influent flume level controller is set up to measure over a 0 to 7 foot range where the desired control set level is approximately 6 feet and the control action is such that the output will swing from 0 to 100% when the level swings from minus 6 inches to plus 6 inches (control band) around the 6-foot level set point. This correlates to a gain of approximately 7, or a proportional band of approximately 14%. The bias constant input is selected to shift the control band to the 6xe2x80x9cnullxe2x80x9d point of the controller. With this configuration, although very stable, flume level is maintained at set point only at the xe2x80x9cnullxe2x80x9d point of the controller. As such, for all practical purposes, an offset between actual flume level and desired set point level always exists. The output (cascaded to the set point inputs of the filter flow controllers) is a predetermined value based on gain (or proportional band) and has an indirect relationship to what is actually flowing into the filters.
FIG. 5 Prior Art is a depiction of the flow into the flume or filters (waveform xe2x80x9cFxe2x80x9d) as contrasted to the calculated output (or in other words the total filter effluent flow set point) over time for a cascaded direct acting proportional only flume level controller with constant bias. This controller output (waveform xe2x80x9cGxe2x80x9d) is cascaded to multiple reverse acting proportional/integral/derivative (PID) filter effluent flow controllers and is in actuality the filter effluent flow set point. For simplicity, it is assumed that the responses of the filter effluent flow PID controllers are instantaneous and that filter effluent flow rate is also depicted by waveform xe2x80x9cGxe2x80x9d. The xe2x80x9cDelayxe2x80x9d depicted by the two vertical dotted lines shows the time it takes subsequent to a change in filter influent flow before the controller begins to make a correction due to level. The thickened horizontal portions of the waveforms at the beginning and at the end of the time period depict where both waveforms xe2x80x9cFxe2x80x9dand xe2x80x9cGxe2x80x9dare resting on top of each other. In this example, it can be seen that xe2x80x9cGxe2x80x9dincreases to a lesser degree than xe2x80x9cFxe2x80x9d(based on typical gain values chosen) and after an initial process delay. This initial delay is proportional to the time it takes the flume level to increase due to the increase in filter influent flow, waveform xe2x80x9cFxe2x80x9d.
FIG. 6 Prior Art is a depiction of the level in the flume or filters over time (waveform xe2x80x9cHxe2x80x9d) as a result of the increased flow into the flume or filters (waveform xe2x80x9cFxe2x80x9dof FIG. 5 previously) as contrasted to the flume level set point (waveform xe2x80x9cIxe2x80x9d). Note that the xe2x80x9cDelayxe2x80x9d depicted by the two vertical dotted lines shows the time it takes subsequent to a change in influent flow before the level in the flume or filters begins to change, and such delay is aligned with the delay described previously for FIG. 5. The thickened horizontal portions of the waveforms at the beginning and at the end of the time period depict where both waveforms xe2x80x9cHxe2x80x9d and xe2x80x9cIxe2x80x9d are resting on top of each other.
The reason the type of level control described for FIG. 3, FIG. 4, FIG. 5, and FIG. 6 is not ultimately satisfactory is because the flow out of the filters is not a direct function of the flow into the filters across the full range of flow. Thus it is demonstrated that only when filter influent flow is at the xe2x80x9cnullxe2x80x9d point as described previously is flume level able to be controlled at set point level.
Another control scheme for flume level control is depicted on FIG. 7 Prior Art. This configuration addresses the offset problem described for FIG. 3, FIG. 4, FIG. 5, and FIG. 6 by upgrading the flume level controller from proportional only (P) to proportional/integral/derivative (PID). In this manner, flume level can be maintained at set point level over the full range of flow into the flume. This is so because the output of the flume level PID controller xe2x80x9csearchesxe2x80x9d for the correct flow set point output to be cascaded to the filter flow PID controllers over time. However, due to the potential for oscillation typical of cascading two PID controllers, both the flume level PID controller and the filter flow PID controllers must be detuned in order to minimize this problem. As a result, the control system is slow to respond and process flow and level fluctuate (overshoot and undershoot) unnecessarily when there is a change in system effluent demand, during a filter backwash, or when filter influent flow changes.
FIG. 8 Prior Art is a depiction of the flow into the flume or filters (waveform xe2x80x9cJxe2x80x9d) as contrasted to the calculated output (or in other words the total filter effluent flow set point) over time for a cascaded direct acting proportional/integral/derivative flume level controller. This controller output (waveform xe2x80x9cKxe2x80x9d) is cascaded to multiple reverse acting proportional/integral/derivative (PID) filter effluent flow controllers and is in actuality the filter effluent flow set point. For simplicity, it is assumed that the responses of the filter effluent flow PID controllers are instantaneous and that filter effluent flow rate is also depicted by waveform xe2x80x9cKxe2x80x9d. The xe2x80x9cDelayxe2x80x9d depicted by the two vertical dotted lines shows the time it takes subsequent to a change in filter influent flow before the controller begins to make a correction due to level. The thickened horizontal portion of the waveform at the beginning of the time period depicts where both waveforms xe2x80x9cJxe2x80x9d and xe2x80x9cKxe2x80x9d are resting on top of each other.
FIG. 9 Prior Art is a depiction of the level in the flume or filters over time (waveform xe2x80x9cLxe2x80x9d) as a result of the increased flow into the flume or filters (waveform xe2x80x9cJxe2x80x9d of FIG. 8 previously) as contrasted to the flume level set point (waveform xe2x80x9cMxe2x80x9d). Note that the xe2x80x9cDelayxe2x80x9d depicted by the two vertical dotted lines shows the time it takes subsequent to a change in filter influent flow before the level in the flume or filters begins to change. The delay described previously for FIG. 8 is longer than this delay due to the PID flume level controller detuning necessary to reduce the effect of oscillation. The thickened horizontal portion of the waveform at the beginning of the time period depicts where both waveforms xe2x80x9cLxe2x80x9d and xe2x80x9cMxe2x80x9d are resting on top of each other. Subsequent delays are experienced as the flume level changes above and below the set point until the controller output xe2x80x9chones inxe2x80x9d on the value of filter effluent flow that gradually brings flume level back to set point.
The reason the type of level control described for FIG. 7, FIG. 8, and FIG. 9 is not ultimately satisfactory is because the flow out of the filters is not a direct function of the flow into the filters across the full range of flow. As can be seen, any change in filter influent flow causes unnecessary modulation in filter effluent flow and flume level in order to bring flume level back to set point level.
The implementation of clearwell level control is first depicted in the configuration shown on FIG. 10 Prior Art. A reverse acting proportional only (P) clearwell level controller with constant bias is cascaded through a low select function to a reverse acting proportional/integral/derivative (PID) plant influent flow controller. The output of the low select function serves as the set point input to the plant influent flow controller. The output of the plant influent flow controller is a position demand signal to the respective valve motor positioner. The output of the low select function is the lesser of the operator adjusted plant influent flow set point (FSP) or the clearwell level controller output. This serves to override the operator adjusted plant influent flow set point if the clearwell level is exceedingly high so that plant influent flow is reduced below the manual influent flow set point. Plant effluent closed loop control is shown with a reverse acting proportional/integral/derivative (PID) system pressure controller whose output is a speed demand to a variable speed drive for a pump.
The control configuration shown in FIG. 10 is not ultimately satisfactory because the flow into the clearwell is not a direct function of the flow out of the clearwell. To minimize this design problem, operations personnel must constantly adjust the plant influent flow set point to match changes in plant effluent demand to keep the clearwell level within high and low limits. This is exceedingly difficult, since the effect of a change in plant influent flow on clearwell level is not seen and therefore cannot be analyzed as to its accuracy until some time after the adjustment is made. This difficult situation is made nearly impossible when adjustments are made during periods when plant effluent flow is also changing to keep up with system demand. Further, the offset problem as described previously for FIG. 3, FIG. 4, FIG. 5, and FIG. 6 with proportional only level control is experienced here as well.
An alternate clearwell level control configuration is depicted on FIG. 11 Prior Art. This is very similar to that described for FIG. 10 except that an added level of control is implemented through a second clearwell level controller and associated low select function. The reverse acting proportional only (P) level controller and low select function both labeled xe2x80x9cNo. 1xe2x80x9d operate as described for FIG. 10 previously. The output of an additional direct acting proportional only (P) clearwell level controller xe2x80x9cNo. 2xe2x80x9d with constant bias, or the output of the plant effluent reverse acting proportional/integral/derivative (PID) system pressure controller is selected through a low select xe2x80x9cNo. 2xe2x80x9d function. The output of low select function xe2x80x9cNo. 2xe2x80x9d is the lesser of these two signals and serves to reduce the speed of the variable speed drive if the clearwell level is exceedingly low.
The control configuration shown in FIG. 11 is not ultimately satisfactory for the same reasons described for FIG. 10. The same operational drawbacks described for FIG. 10 also apply here.
The clearwell level control configuration depicted on FIG. 12 is a modification to that described in FIG. 10. It removes the low select function and addresses the offset problem described for FIG. 3, FIG. 4, FIG. 5 and FIG. 6 by upgrading the clearwell level controller from proportional only (P) to proportional/integral/derivative (PID). In this manner, clearwell level can be maintained at set point level over the full range of flow into the plant. This is so because the output of the clearwell level PID controller xe2x80x9csearchesxe2x80x9d for the correct flow set point output to be cascaded to the plant influent flow PID controller over time. However, due to the potential for oscillation typical of cascading two PID controllers, both the clearwell level PID controller and the plant influent flow PID controller must be detuned in order to minimize this problem. As a result, the control system is slow to respond and process flow and level fluctuate (overshoot and undershoot) unnecessarily when there is a change in system effluent demand or during a filter backwash.
The clearwell level control configuration depicted on FIG. 13 is a modification to that described in FIG. 12. It replaces the proportional/integral/derivative (PID) clearwell level controller with a proportional only (P) controller with variable bias. The variable bias is the desired plant influent flow set point. Thus, when clearwell level is at desired set point, the level controller output to the set point input of the plant influent flow controller is the manually adjusted flow set point, FSP. This is an improvement over the first design described in FIG. 10 in that operations personnel can be less attentive to making adjustments to the plant influent flow set point, FSP. However again, this control configuration shown in is not ultimately satisfactory because the flow into the clearwell is not a direct function of the flow out of the clearwell. Further, the offset problem as described previously for FIG. 3, FIG. 4, FIG. 5, and FIG. 6 with proportional only level control is experienced here as well.
FIG. 14 Prior Art is a depiction of the flow into the clearwell (i.e. out of the filters) as contrasted to the effective total plant effluent flow over time for a cascaded direct acting proportional/integral/derivative (PID) flume level controller in conjunction with a cascaded reverse acting proportional/integral/derivative (PID) clearwell level controller. The thickened portion of the waveforms shown at the far left up to the dotted vertical line denoting the beginning of the xe2x80x9cBackwash Sequence (Filter Off Line)xe2x80x9d shows waveforms xe2x80x9cNxe2x80x9d, xe2x80x9cOxe2x80x9d, xe2x80x9cPxe2x80x9d, and xe2x80x9cQxe2x80x9dlaying on top of each other and an otherwise steady state condition. When the backwash sequence begins and the pertinent filter is taken off line, waveform xe2x80x9cOxe2x80x9ddepicting total filter effluent flow drops briefly below waveform xe2x80x9cNxe2x80x9d. Waveform xe2x80x9cNxe2x80x9d depicts effective total plant effluent flow out of the clearwell as well as the PID flume level controller output which acts as the set point input to the filter effluent PID controllers. Partway into the xe2x80x9cBackwash Sequence (Filter Off Line)xe2x80x9d period but before the beginning of the xe2x80x9cWashwater Flow Periodxe2x80x9d waveform xe2x80x9cOxe2x80x9d rejoins waveform xe2x80x9cNxe2x80x9dat the point where it splits into waveform xe2x80x9cPxe2x80x9dand waveform xe2x80x9cQxe2x80x9d. Waveform xe2x80x9cPxe2x80x9ddepicts the PID flume level controller output (acting as the set point input to the filter effluent flow PID controllers) and for simplicity, it is assumed that the responses of the filter effluent flow PID controllers are instantaneous and that this waveform depicts filter effluent flow as well. Waveform xe2x80x9cQxe2x80x9d depicts the effective plant effluent flow output and demonstrates increased clearwell outflow due to pumped backwash water flow during the xe2x80x9cWashwater Flow Periodxe2x80x9d. The waveforms shown on this figure are time synchronized and related to the waveforms shown on FIG. 15 and FIG. 16.
FIG. 15 Prior Art is a depiction of the level in the flume or filters over time (waveform xe2x80x9cRxe2x80x9d) as contrasted to the flume level set point (waveform xe2x80x9cSxe2x80x9d) as a result of a cascaded direct acting proportional/integral/derivative (PID) flume level controller in conjunction with a cascaded reverse acting proportional/integral/derivative (PID) clearwell level controller. In particular, waveform xe2x80x9cRxe2x80x9d is the response of the flume/filter level as a result of the disturbance placed on the system by the start of the backwash sequence depicted in FIG. 14. The thickened horizontal portion of the waveform at the beginning of the time period depicts where both waveforms xe2x80x9cRxe2x80x9d and xe2x80x9cSxe2x80x9d are resting on top of each other. The waveforms shown on this figure are time synchronized and related to the waveforms shown on FIG. 14 and FIG. 16.
FIG. 16 Prior Art is a depiction of the level in the clearwell over time (waveform xe2x80x9cTxe2x80x9d) as contrasted to the clearwell level set point (waveform xe2x80x9cUxe2x80x9d) as a result of a cascaded direct acting proportional/integral/derivative (PID) flume level controller in conjunction with a cascaded reverse acting proportional/integral/derivative (PID) clearwell level controller. In particular, waveform xe2x80x9cTxe2x80x9dis the response of the clearwell level as a result of the disturbance placed on the system by the start of the backwash sequence and the start of the washwater flow period depicted in FIG. 14. The thickened horizontal portion of the waveform at the beginning of the time period depicts where both waveforms xe2x80x9cTxe2x80x9dand xe2x80x9cUxe2x80x9d are resting on top of each other. The waveforms shown on this figure are time synchronized and related to the waveforms shown on FIG. 14 and FIG. 15.
FIG. 17 Prior Art is a replication of FIG. 14 for the purpose of showing time synchronization and interrelationship of numerous system variables depicted as waveforms on FIG. 18 and FIG. 19.
FIG. 18 Prior Art is a replication of FIG. 15 and FIG. 16 for the purpose of showing time synchronization and interrelationship of numerous system variables depicted as waveforms on FIG. 17 and FIG. 19.
FIG. 19 Prior Art is a depiction of a typical filter effluent flow proportional/integral/derivative controller output (waveform xe2x80x9cVxe2x80x9d) as contrasted to a steady state baseline (waveform xe2x80x9cWxe2x80x9d) as a result of a cascaded direct acting proportional/integral/derivative (PID) flume level controller in conjunction with a cascaded reverse acting proportional/integral/derivative (PID) clearwell level controller. In particular, waveform xe2x80x9cVxe2x80x9dis the position demand output to the filter effluent valve positioners (for the remaining filters not being backwashed) in response to the flume level as a result of the disturbance placed on the system by the start of the backwash sequence and the start of the washwater flow period depicted in FIG. 17. The thickened horizontal portion of the waveform at the beginning of the time period depicts where both waveforms xe2x80x9cVxe2x80x9d and xe2x80x9cWxe2x80x9d are resting on top of each other. The waveforms shown on this figure are time synchronized and related to the waveforms shown on FIG. 17 and FIG. 18. Additionally, waveform xe2x80x9cXxe2x80x9d is the speed demand output to the variable speed pump drive for the system pressure controller (while waveform xe2x80x9cYxe2x80x9d is a steady state baseline) as a result of the change in clearwell level due to the disturbance placed on the system by the start of the backwash sequence and the start of the washwater flow period depicted in FIG. 17. The thickened horizontal portion of the waveform at the beginning of the time period depicts where both waveforms xe2x80x9cXxe2x80x9d and xe2x80x9cYxe2x80x9d are resting on top of each other.
As can be seen, the figures show a certain amount of instability as well as oscillatory dynamics built into the designs of the prior art. This results in excessive wear and tear on flow regulating devices such as valves and pumps. For the water treatment industry, these designs do not minimize the potential for turbidity breakthrough, and therefore do not provide maximum treatment potential.
The object of this invention is to provide a fluid transportation system which has solved the previously described problems of the prior art. The measured levels of a system of a plurality of vessels or reservoirs interconnected by pipeline or open channel in which flow measuring and manipulating devices are placed can be maintained at desired level set points by appropriately regulating such flows. For a given number of reservoirs or vessels in a system where demand responsiveness is required of the inflow to the first vessel or reservoir in a series, outflow regulation is accomplished for all vessels or reservoirs in the series utilizing as many replications as necessary of a module for each vessel or reservoir consisting of a level controller whose output is cascaded forward to the flow controller(s) for the flow out of such vessel or reservoir in the series, this control module labeled as the xe2x80x9cnestedxe2x80x9d module.
All level controllers are configured as proportional only with variable bias resulting from universal equations such that the controller is immediately responsive to changes in system flow in advance of an actual change in level. The variable bias calculation contemplates universal system dynamics such as whether the flow controllers receiving the associated level controller output as their respective set point inputs are actually able to be responsive to such set point input.
As a result of the above, ultimate system responsiveness and stability is realized, levels are maintained at desired set points with minimum delay and with no oscillation even in the most significant of system disturbances. Because of this, the wear and tear on the flow manipulating devices such as valves and pumps is minimized. Also, flow rate fluctuations are minimized, and water quality maximized. Finally, the inherent inaccuracies associated with any associated data collection system are minimized because the invention reveals calibration errors and discrepancies amongst the various flow-measuring devices that typically go undetected for indeterminate amounts of time.
Further objects and advantages of this invention will become apparent from a consideration of the drawings and ensuing description.